In conventional read channel integrated circuits offsets caused by the elements of the read channel integrated circuits introduce errors into the signal. To offset these errors autozeroing techniques have been used which zero the analog read signal applied to the signal path at a first location in the read channel so that the signal in the path is now the undesired offset. Then a signal at a location downstream from the first location is fed to a storage device which is coupled to a point between the first and second locations to cancel the offset. The storage device is then decoupled from the second location and the zeroing of the read signal is ended. Such devices are disclosed in U.S. Pat. Nos. 5,648,738 (disc drives) and 5,508,656 (tape drives).
Separately, another problem is that in such read channels the input to the analog to digital converter (ADC) is preferably large but the output from the preceding filter is typically small. If the output was made larger such as by a prior amplification then the filter output would contain more distortion. Placing an amplifier between the filter and ADC adds another set of problems. If that amplifier introduces substantial amplification then the bandwidth will be narrowed and harmonic distortion will be increased. If the gain is decreased then bandwidth can be broader and harmonic distortion lower but the signal to noise ratio will be poorer at the input of the ADC.
One approach is to simply place a capacitor in series between the filter and ADC but that capacitor must be very large in order to block the d.c. offset yet pass the high frequency signal. Such a capacitor would be too large for implementation in an integrated circuit chip.